J. Am. Chem. SOC.1994, 116, 5196-5205
5196
Synthesis, Structure, and Reactivity of (p-Oxo)bis( p-carboxylato)diiron( 111) Complexes of a Dinucleating Dicarboxylate Ligand, Stable Models for Non-Heme Diiron Protein Cores Stephen P. Watton, Axel Masschelein, Julius Rebek, Jr., and Stephen J. Lippard' Contribution from the Department of Chemistry, Massachusetts Institute of Technology, Cambridge, Massachusetts 021 39 Received January 14, 1994"
Abstract: Use of a pre-organized, cleft-shaped dicarboxylate ligand, m-xylenediamine-bis(Kemp'striacid)imide (XDK), in the self-assembly synthesis of (p-oxo)bis(p-carboxylato)diiron(III) complexes as models for the active site of nonheme diiron protein cores resulted in exclusive formation of dinuclear complexes. Reaction of HzXDK with Fe(NO3)~-9H20in methanol solution afforded multigram quantities of a (p-oxo)bis(p-carboxylato)diiron(III) complex having six solvent ligands in the terminal positions. The geometry of this complex, [Fe20(XDK)(CH30H)5(H20)](N03)2*4H20 ( 1 (NO3)2*4H20),was determined by X-ray crystallography, and Mossbauer spectroscopy demonstrated the persistence of its dinuclear structure in frozen methanol solution. Compound 1 reacted cleanly with N-donor ligands such as 2,2'-dipyridyl (bpy) to afford terminally substituted products, for example [Fe20(XDK)(bpy)2(CH30H)2]2+ (2). The reactions were readily monitored by changes in the optical spectroscopic features of the complexes, and a stopped-flow kinetics analysis of the ligand exchange reaction revealed the second order rate constant k to be (1.51 f 0.04) X lo3 M-l s-l. This value is quite similar to the rate constants for ligand substitution at the terminal site in hemerythrin. The substitution by two bpy ligands, one on each iron of the dinuclear center, proceeds at the same rate, except for a statistical factor of 2 favoring binding of the first equivalent, so that k = k2 = 0.5kl. Various solid products wereobtained from this reaction, [ Fe20(XDK)(bpy)2XY]"+,depending upon the isolation prccedureused. Crystallization from a methanol/diethyl ether mixture gave the unsymmetrical complex 3 (X = CH30H, Y = NO3-, n = 1); recrystallization from CH2Cl2/toluene gave the symmetrical anti-4 (X = Y = NO3-, n = 0); and recrystallization from acetonitrile afforded syn-4 (X = Y = NO3-, n = 0), the X-ray structure of which is reported. Reaction of l(N03)2 with 1-methylimidazole (1-MeIm) in methanol solution resulted in successive replacement of each of the terminal solvent ligands, affording [Fe20(XDK)( l-MeIm)6]2+ (5). Formation of this complex was monitored spectroscopically with the use of UV/visible and resonance Raman spectroscopy, and its solid state structure was determined by X-ray crystallographic analysis of the tetraphenylborate salt.
Introduction The occurrence of carboxylate-bridged diiron centers at the active sites of non-heme iron proteins such as hemerythrin (Hr), ribonucleotide reductase (RR), purple acid phosphatase (PAP), and methane monooxygenase (MMO) has stimulated extensive study of low molecular weight models for their dimetallic~ores.~-~ Although a large number of structurally and spectroscopically relevant model complexes for both the reduced and oxidized forms of the protein active sites now exist, the biological activities have not yet been effectively reproduced. One difficulty associated with achieving biomimetic reactions with non-heme iron-containing model complexes stems from rapid substitution kinetics at the metal ions, which afford access to multiple binding modes and a variety of oligomeric products. This situation contrasts with the behavior of the protein-bound non-heme iron centers, where the three-dimensional folding of the polypeptide chain imposes thermodynamic and kinetic restrictions on the active site chemistry. The result is specific and controlled reactivity. In order to provide better control over the reactivity of carboxylate-bridged diiron model complexes, with the ultimate aim of reproducing the functions of metalloproteins and enzymes e Abstract
published in Advance ACS Abstracts, May 1, 1994. (1) Lippard, S.J. Angew. Chem., Int. Ed. Engl. 1988, 27, 344-361. (2) Sanders-Loehr, J.; Wheeler, W. D.; Shiemke, A. K.; Averill, B. A.; Loehr, T. M. J. Am. Chem. SOC.1989, 111, 8084-8093. (3) Kurtz, D. M., Jr. Chem. Rev. 1990, 90, 585-606. (4) Vincent, J. B.; Olivier-Lilley, G. L.; Averill, B. A. Chem. Reu. 1990, 90, 1447-1467. (5) Que, L., Jr.; True, A. E. h o g . Inorg. Chem. 1990, 38, 97-200.
containing these units, we have been exploring the extent to which the principles of molecular recognition in organic chemistry might be applied to bioinorganic modeling studies. Specifically, we have employed the cleft-shaped dicarboxylate ligand xylenediamine-bis(Kemp's triacid)imide,6v7hereafter XDK, to assemble
XDKH,
a dinuclear (Fe20(XDK))2+ core structure having six labile solvent0 ligands in the terminal positions, [Fe20(XDK)(CH3OH),(H20)]2+ ( 1 ) . This remarkable complex is resistant to oligomerization, existing entirely as the dinuclear species in solution. Moreover, compound 1 readily converts to a variety of (oxo)bis(carboxylato)-bridged dinuclear Fe( 111) derivatives, and in this respect it is superior to our previous attempt to use a dinucleating dicarboxylate ligand to stabilize the (Fe20(02CR)2)2+ core.* In the present article we describe the synthesis, (6) Rebek, J., Jr.; Marshall, L.; Wolak, R.; Parris, K.; Killoran, M.;Askew, B.; Nemeth, D.: Islam, N. J . Am. Chem. SOC.1985, 107, 7476-7481. (7) Marshall, L.; Parris. K.: Rebek, J.. Jr.: Luis, S.V.: Burnuete, - M. I. J. Am. Chem. SOC.1988, 110, 5192-5193. (8) Beer, R. H.; Tolman, W. B.; Bott, S.G.; Lippard, S.J. Inorg. Chem. 1991, 30, 2082-2092.
0002-7863/94/1516-5196$04.50/00 1994 American Chemical Society
(p-Oxo)bis(p-carboxylato)diiron(III)Complexes
J. Am. Chem. SOC.,Vol. 116, No. 12, 1994 5 191
characterization, and terminal ligand substitution reactions of 1, the last of which should be of value for understanding ligand exchange processes taking place during turnover in t h e enzymatic systems. T h e utility of 1 in preparing a model for t h e dinuclear iron center and nearby tyrosyl radical in the RR R2 protein was previously communicated.9 Dinuclear iron(I1) complexesof XDK have been independently investigated,but in the context of capping rather than skeletal ligation.10
except that acetonitrile was employed in the recrystallizations, affording the product as green blocks. Metinnil. Reactionof0.100g (172pmol)ofH2XDK,O.l18mg(172 pmol) of [Fe(bpy)(N03)2]20,and 46 pL of (CzH5)sN in 10 mL of CH3OH afforded 0.100 g (51%) after recrystallization from 10 mL of acetonitrile. MetliodZ. A 0.085-g (0.227 mmol) portion of 3(NO3) was recrystallized from 8 mL of acetonitrile to afford 0.050 g (60.4%) of syn-4. Anal. Calcd for C S ~ H ~ ~ N ~ OC,I 54.66; ~ F ~ ZH,: 4.76; N, 9.81. Found: C, 54.15; H, 4.75; N, 10.18. FTIR (thin film on KBr plate): 2962, 2926, Experimental Section 1732, 1690, 1654, 1541, 1405, 1356, 1230, 1197, 1023, 968, 835,768, 736,695 cm-I. UV/vis (CH3CN) Am, nm (e, Fe-I, M-1 cm-I): 332 sh Materials and Methods. HzXDK was prepared as described previ(3707), 366 sh (2375), 482 (432), 528 sh (143), 684 (100). 0usly,6-~except that the 2,4-diamino-m-xylene starting material was [FQO(XDK)(I-M~I~)~](BP~)+CHCI~, S ( B ~ ) z C H C I 3The comobtained by an alternative, more convenient procedure, involving nitration plex was prepared by treatment of 1(N03)*(0.2 g, 0.19 mmol) with an of m-xylene followed by fractional crystallization and catalytic hydroexcess of 1-MeIm (0.87 mL, 57 equiv) in methanol (2 mL) to give a deep genation of the 2,5-dinitroxylene product. The complex [Fe(bpy)green-brown solution. A solution of N a B P b (0.25 g, 0.73 mmol) in (N03)2]20 was also prepared by a literature procedure.Il All reagents methanol (2 mL) was added, affording a pale green precipitate, which were purchased from commercial sources and used as received unless was collected by filtration and dried (yield 0.34 g, 96%). The complex otherwise noted. was recrystallized by vapor diffusion of hexane into a concentrated Preparation of Compounds. [F~~~(XDK)(CH~OH)~(HZO)](NO~)~ chloroform solution. Anal. Calcd for C I W H I I ~ N I ~ O ~ FC, ~ Z67.98; BZ: 4 H B , 1(N&)r4Hfi. This complex was prepared by treating a H, 6.25; N, 10.67. Found: C, 67.43; H, 6.29; N, 10.72. FTIR (KBr suspension of HzXDK (5.0g, 8.6 mmol) in methanol (100 mL) with pellet) 3134, 3051, 2978, 2921, 1729, 1658, 1540, 1469, 1424, 1368, Fe(NO3)~9H20(6.87 g, 17 mmol) to give a deep green solution. Diethyl 1275, 1226, 1191, 1099, 1026, 949,849,736,701,662, 614, 501 cm-l. ether was added to bring the volume to 250 mL, the mixture was filtered, UV/vis (CHClp) ,A,, nm (e, Fe-', M-I cm-I): 330 sh (2147), 364 sh and the filtratewascooled to-20 OCovernight. Thecomplexwasobtained 505 (1796), 480 (271), 680 (49). Raman (CH30H): u,,(Fe-O-Fe) as green blocks which were collected and dried in air. Yield 6.8 g, 73%. cm-l . Calcd for C37H68N4025Fez: C, 41.12; H, 6.34; N, 5.18. Found: Anal. X-ray Crystallography. Data collection and reduction, including C, 41.43; H, 5.74; N, 5.63. FTIR (KBr pellet): 3400 (br), 3196 (br), corrections for Lorentz, polarization, and absorption effects, were 2969,1733, 1671,1543,1470,1415, 1372,1295,1202,1095,1021,961, performed by using general procedures previously described.lZ No 885,846,773,634,519 cm-I. UV/vis (CHaOH),, ,A nm (e, Fe-l, M-1 appreciable decay was observed for any of the compounds, as judged by cm-1): 358 (1656), 474 (157), 616 (57). Raman (CHsOH): v,,(Fe periodic monitoringof theintensitiesof threestandardreflections. Initial 0-Fe) = 532 cm-I. MGssbauer (solid state): Site 1, 8 = 0.55 mm s-l, iron positions were obtained by using the direct methods programs AEQ = 1.67 mm s-l. Site 2, d = 0.56 mm s-l, AEQ = 2.06 mm s-I; MITHRIL and SHELXS-86I3 for 1, 4, and 5. The remaining heavy (CH30H, frozen solution): 8 = 0.56 mm s-I, AEQ = 1.87 mm s-I. atoms were located with DIRDIF phase refinements and difference Fourier [FezO(XDK)(bpy)z(CH3OH) (N03)]( N o d , 3(NO3). A portion of maps.14 The TEXSAN package of programs was used to refine the [Fe2O(XDK)(CH~OH)~(H20)](N03)~(0.25Og,0.227 mmol) wasstirred structure^.^^ A summary of data collection parameters for complexes 1, in 1 mL of methanol to give a deep green solution. Solid 2,2'-dipyridine 4, and 5 is given in Table 1. (0.071 g. 0.454 mmol) was added and the resulting brown solution was [FezO(XDK)(CH~OH)~(HZO)](NO~)~~H~O, 1( N O J ) ~ . ~ H PThe . stirred for 5 min and then filtered. Vapor diffusion of ether into this crystal used in the X-ray diffraction study, obtained directly from the solution at room temperature afforded the product as brown blocks. Yield, synthetic procedure described above (dimensions 0.4 X 0.4 X 0.5 mm), 0.285 g (96%). Anal. Calcd for C53H58Ng016Fe~:C, 54.19; H, 4.98; was mounted in Paratone N on the end of a quartz fiber and frozen at N, 9.54. Found: C, 53.83; H, 5.02; N, 9.54. FAB-MS (3-NBA matrix): 195 f 1 K. Data collection and structure solution were performed as m / e 1080 (M+ - CHsOH), 1018 (1080 - NOo), 924 (1080 - bpy), 862 described above. The positions of all atoms were refined anisotropically, (1018 - bpy). FTIR (KBr pellet): 3419 (br), 3264 (br, sh), 2978,2920, with theexception of thelatticesolvents, which were refined with isotropic 1731,1686, 1605,1549,1453,1368,1325,1288,1190,1104,1029,964, thermal parameters. The hydrogen atoms bound to 01, 0 2 , and 0 3 892,848,774,733,659,634cm-I. UV/vis (CH30H), ,A,, nm (e, Fe-I, were located from difference Fourier maps, assigned isotropic thermal M-l cm-1): 330 sh (4145), 364 sh (2702), 461 (439), 490 (440), 528 sh parameters, and refined. The carbon-bound hydrogen atoms were placed ( l l l ) , 694 (70). at calculated positions in the final refinement cycles. The hydrogen atom anti-[Fefl(XDK)(bpy)~(NO3)zl, a n t i 4 Method 1. The compounds bound to 0 4 was not located. The largest peak in the final difference HzXDK (0.100 g, 172pmol) and [Fe(bpy)(NO&]zO (0.1 18g, 172pmol) Fourier map was approximately 0.52 e-/A3, located in the vicinity of the were stirred in 10 mL of methanol, and Et3N (46 pL, 350 pmol) was lattice solvents. added. The resulting deep green-brown solution was filtered to remove syn{Fe@( XDK) (bpy)z(NO3)$CH3CN, sp43CHjCN. The crystal asmallamount ofredmaterial, thenether (50mL) wasadded toprecipitate used in the X-ray study was obtained by recrystallization at room the product. The pale green-brown solid was collected and dried (160 temperature, as described above (Method 1). A brown-green block mg, 8 1%). The complex was recrystallized fromCHzClz/tolueneas large (dimensions 0.1 X 0.1 X 0.15 mm) was mounted on a glass fiber using dark green blocks. The yield of crystallized material was 0.100 g (50% Paratone N oil. Data were collected at 195 f 1 K with w-28 scans. based on HzXDK). Anal. Calcd for C ~ Z H ~ ~ N ~ OC,I 54.66; ~F~Z H,: Psi-scans for four independent reflections indicated that no correction 4.76; N, 9.81. Found (mean of two samples): C, 54.69 f 0.90; H, 4.72 was needed for absorption effects, the relative intensities varying from f 0.05; N, 9.65 f 0.04. FAB-MS (NBA matrix): m / e 1080 (M+ 0.95 to 1.OO. All non-hydrogen atoms were refined anisotropically, except NO& 1018 (M+ - 2N03), 924 (1080 - bpy), 862 (1018 - bpy). FTIR for those of the lattice acetonitrile molecules, which were refined using (KBr pellet): 3417, 2978, 2934, 1727, 1696, 1553, 1461, 1360, 1190, isotropic temperature factors. Hydrogen atoms were included in the 770, 734 cm-I. UV/vis (CHZClz), ,,A, nm (e, Fe-l, M-I cm-I): 330 final cyclesof least-squares refinement at calculated positions. The largest (5837), 365 (3337), 464 (448), 484 (426), 686 (79). Raman (CHCl3): electron density peak in the final difference Fourier map corresponded v,,(Fe-O-Fe) 532 cm-I. to 0.7 e-/A3, located near a lattice acetonitrile molecule. Metinnit. Compound 3(NO3) (0.100 g, 85 pmol) was recrystallized [F@(XDK)( l - M e I m ) ~ ] ( B ~ ) z - C H5(BP~)2CHCl3 Cl~ The crystal from CHZClZ/toluene to give 0.07 g (75%) of green crystals, identical used in the X-ray study was grown by vapor diffusion of hexane into a by IR spectroscopy with the product prepared by Method 1. syn-[FezO(XDK)(bpy)2(NO3)21, syn-4, The i y n isomer of 4 was prepared by using the methods described for the anti isomer (above), (9) Goldberg, D. P.; Watton, S. P.; Masschelein,A.; Wimmer, L.; Lippard, S . J. J . Am. Chem. SOC.1993. 115. 5346-5341. ( 10) Hagen, K. S.; Lachicotte, R.; Kitaygorodskiy,A.; Elbouadili, A. Angew. Chem., Znt. Ed. Engl. 1993, 32, 1321-1324. (1 1) Taft,K. L.;Masschelein,A.;Liu,S.;Lippard,S.J.;Garfinkel-Shweky, D.; Bino, A. Znorg. Chim. Acta 1992, 198-200,621-631.
(12) Carnahan, E. M.; Rardin, R. L.; Bott, S. G.; Lippard, S. J. Znorg. Chem. 1992, 31, 5193-5201. (13) Sheldrick, G. M. In Crystallographic Computing$ Sheldrick, G. M.,
KrBger, C., Goddard, R., Eds.; Oxford University Press: Oxford, 1985; pp 175-189. (14) Parthasarathi, V.; Beurskens, P. T.; Slot, H. J. B. Acta Crystallogr. 1983, A39, 86G-864. ( 15) TEXSAN: Single Crystal Structure Analysis Software, Version I .6; Molecular Structure Corporation: The Woodlands, TX, 1992.
5198 J. Am. Chem. SOC.,Vol. 116, No. 12, 1994
formula fw
crystal system space group a, A
b, A c, A
CnH68N4O25Fe2 1080.7 orthorhombic
CsnHaNiiOI& 1265.9 monoclinic
CiosHi isNi409B2C13Fe2 1956.8 triclinic
Pbcm
Ri/n
Pi
11.736(3) 20.635(7) 22.223(7)
13.462(4) 22.632(4) 20.002(7)
15.133(3) 23.798(3) 14.468(2) 1OO.74(1) 93.68(1) 101.26(1) 4993( 1)
a,deg
92.92(2)
8, deg 7 , deg
v,A3
Z d,l,, g cm-' T,K data coll range, deg data limits no. of data collected p-factorb R." no. of obs unique datae no. of parameters data/parameter ratio abs coeff, cm-I trans. coeff, min/max Rd R, largest shift/esd, final largest peak, e-/A'
5382(3) 4 1.334 195 3 d 29 d 42
6086(3) 4 1.381 195 3 d 29 5 48
+h,+k,+l
+h,+k,il
+h,fk,fl
4852 0.05 0.020 2098 326 6.44 6.19 0.887/1 .OOO 0.068 0.089 0.000 0.524
9793 0.03 0.040 5539 175 7.14 5.51 0.945/1.000 0.056 0.066
13805 0.022 0.033 7462 96 1 7.76 4.34 0.934/ 1.000 0.056 0.059 0.000 0.726
0.000
0.722
2
1.301 20 1 3 S 29 5 45
Watton et al. containing species at equilibrium in solution.lJG1* Kinetically rapid substitution reactions at the metal centers compound the problem by increasing the accessibility of energetically similar coordination modes. As a consequence, the success of the commonly used "self-assembly" approach is highly sensitive to the reaction conditions. The products of dissociation or oligomerization of the dinuclear complexes are frequently obtained rather than, or in addition to, the desired dinuclear species. For example, the first syntheses of dinuclear (p-oxo)bis(p-carboxylato)diiron(III) models employed the facially capping tridentate ligands [HB(pyrazolyl)3]- and 1,4,7-triazacyclononane(TACN) to preclude the formation of oligomers. Although oligomeric products were largely avoided by this tactic, stable mononuclear [FeL2] complexes were obtained in addition to the desired dinuclear s p e c i e ~ . ~ ~The , ~ Ostructural and spectroscopicfeatures of these initial dinuclear complexes were in excellent agreement with those of the oxidized forms of Hr and RR. These model complexes were unable to reproduce key functional features of the dinuclear iron protein cores, however, since they lack open or readily exchangeable terminal coordination sites that serve in the metalloproteins as sites for reactions with dioxygen or other exogenous ligands. When the tridentate capping ligands are replaced with analogous bidentate donors in self-assembly syntheses, oligomerization of the dinuclear core commonly occurred through formation of additional bonds to the bridging oxo group. The resulting dimer-tetramer equilibrium is shown schematically in eq 1.16J1 In the absence of multidentate capping ligands, bridging through the terminal positions can take place, giving rise to higher nuclearity products.22
a Data were collected on an Enraf-Nonius CAD4-F kappa geometry diffractometerusing Mo K a radiation. Used in thecalculation of a(F). Observation criterion I > 3 4 ) . R = XllF,,l- IFoll/XIFd, R, = [Zw(lFol - IFcl)2/~wIFo12]1/2, where w = 1/u2(F), as defined in Carnahan et al., ref 12.
CHC13 solution at room temperature. A dichroic green-brown block (dimensions 0.3 X 0.3 X 0.4 mm) was mounted in Paratone N on the end of a quartz fiber. The data collection temperature was 201 1 K. The data collection and structure solution werecarried out as described above. The positions of the non-hydrogen atoms of the cation were refined with anisotropic thermal parameters. The phenyl rings of the counterions were treated as rigid bodies. The hydrogen atoms were included in calculated positions during the last cycles of the refinement. The largest peak in the final difference map was 1.O e-/A3, located in the vicinity of one of the B P 4 - phenyl rings. Physical Measurements. lH N M R spectra were collected on a Varian VXR-300 instrument. IR spectra were obtained and manipulated by using a Bio-Rad SPC3200 FTIR instrument and UV/visiblespectra were recorded with a Hewlett-Packard diode array spectrophotometer. Resonance Raman data were obtained by using a Kr ion laser (excitation wavelength 406.7 nm), with an approximateincident power on the sample of 20 mW. A 0.6-m single monochromator (1200 grooves/mm grating), with an entrance slit of 300 pm, and an optical multichannel analyzer were employed in a standard backscattering configuration. A holographic Raman edge filter was used to attenuate Rayleigh scattering. Twenty scans were summed and baseline correction was applied. Stopped-flow kinetic measurements were performed with a Hi-tech Scientific SF-40 stopped flow instrument (Canterbury Stopped-Flow), interfaced to an Apple Macintosh IIfx computer using the software package Labview2. Kinetic runs were performed in triplicate, and the data were averaged prior to fitting. Data were processed and plotted using Kaleidagraph. Mdssbauer spectra werecollected at 100 f 10 Kusinga Ranger Scientific instrument, and the experimental data were either fit to Lorentzian or Voight line shapes. All isomer shifts were referenced to iron metal at 300 K.
Results and Discussion Syntheses. Many difficulties encountered in preparing model complexes for carboxylate-bridged dinuclear non-heme iron proteins are associated with the presence of multiple iron-
Adjustment of the reaction conditions to bias the equilibria toward the formation of dinuclear species has afforded several relevant model complexes having the formulation [Fe20(02CR)zBzM2], where B and M represent bidentate and monodentate ligands, respectively. The first reported example of this type of complex, [ Fe20(OAc)z(bpy)2C12],was prepared by cleaving the tetranuclear core of [Fe402(0Ac)7(bpy)2]+ with excess bpy and chloride ion.16 In addition, disruption of trinuclear precursors by addition of ferric ion and bidentate N-donors afforded dinuclear complexes with neutral or anionic donors as the monodentate ligands.Il In a similar but more direct approach, a dinuclear complex was prepared by self-assembly in which the dinucleartetranuclear equilibrium was circumvented by an excess of the monodentate ligand.23 Preparation of [ Fe20(02CR)2B2M2] complexes was also accomplished by reacting carboxylic acids or (16) Vincent, J. B.; Huffman, J. C.; Christou, G.; Li, Q.;Nanny, M. A.; Hendrickson, D. N.; Fong, R. H.; Fish, R. H. J . Am. Chem. SOC.1988,110, 6898-6900. (17) Driieke, S.;Wieghardt, K.; Nuber, B.; Weiss, J.; Fleischhauer, H.-P.; Gehring, S.;Haase, W. J. Am. Chem. SOC.1989, 111, 8622-8631. (18) Kurtz, D. M., Jr. Chem. Rev. 1990, 90, 585-606. (19) Armstrong, W. H.; Spool,A.; Papaefthymiou, G. C.; Frankel, R. B.; Lippard, S. J. J. Am. Chem. SOC.1984, 106, 3653-3667. (20) Wieghardt, K.; Pohl, K.; Gebert, W. Angew. Chem., Int. Ed. Eng. 1983, 22, 727. (21) Gorun, S. M.; Papaefthymiou, G. C.; Frankel, R. B.;Lippard, S.J. J . Am. Chem. SOC.1987, 109, 3337. (22) See, for example: Taft, K. L.; Lippard,S.J. J. Am. Chem. SOC.1990, 112, 9629-9630. (23) Shteinman, A. A. Mendeleeu Commun. 1992, 155-157.
(p- Oxo)bis(p-carboxylato)diiron(III) Complexes
their salts with preformed p-oxodiiron(II1) precursors,' 124,25 and by oxidation of a diferrous precursor.26 An interesting complex was recently described having an unsymmetrical hemerythrinlike substitution pattern.27 In this complex, one iron bears a tridentate (Me3TACN) capping ligand, while the other is capped by dipyridyl and a monodentate ligand L, where L = H20 or C1-. Preparation of this dinuclear complex was accomplished by using mononuclear [FeC13(Me3TACN)] and dipyridyl to disrupt the tetranuclear core of [Fe402(0Ac)7(bpy)~]+.Terminal ligand substitution was demonstrated in this system by converting the chloro to the aqua derivative following treatment with Ag+ ion. To date, there have been no reports of structurally characterized [ Fe202(02CR))~]2+complexes having the terminal FeM3 substitution pattern outside of a protein environment. Spectroscopic evidence suggested the formation of such a species upon methanolysis of basic iron acetate,22but attempts to isolate the oxo-bridged complex resulted in condensation to give the decanuclear molecular ferric wheel, an oligomer of the [Fe(02CCH2Cl)(OCH3)2] unit.22,28 The lackof well-characterized [Fe20(02CR)~(L)6]complexes with exclusively.monodentate terminal ligands is probably again due to multiple equilibria in solution. Since no chelate effect is operative to stabilize the binding of monodentate capping ligands such as pyridine, they will be less tightly bound and consequently less effective at inhibiting oligomerization equilibria in solution. Moreover, the pK,values of protic solvent ligands such as methanol or water are lowered on coordination to iron(III), affording potential bridging ligands, alkoxide or hydroxide. An important tactic in preparing functional models for the carboxylate-bridged dinuclear non-heme iron proteins is therefore to gain control over the chemistry of the terminal coordination sites, since these are the sites where the biological chemistry is generally expressed. The strategy employed here to accomplish this goal has been to use a dinucleating carboxylate ligand which provides a stable, well-defined bridging environment for assembly of the dinuclear center. The ligand XDK has a cleft-like geometry originally intended to mimic the convergent arrangement of functional groups at enzyme active sites. Its preorganized binding site eliminates conformational degrees of freedom and lowers the entropic barrier to formation of metal complexes, resulting in tighter binding. Enhanced affinity for the binding of Ca2+to a close relative of XDK was demonstrated previ~usly.~It was therefore anticipated that binding of the (Fe2014+unit to the carboxylates of XDK might impart some thermodynamic or kinetic stability to the dinuclear unit. The ligand introduces two bridging carboxylate groups in a geometry that is appropriate for binding to the (p-oxo)diiron(III) core. Binding of this unit to the diacid sequesters the metal centers in a shallow cleft, which, due to the steric influence of the peripheral methyl groups, is quite effective at blocking the formation (eq 1) of head-to-head dimers.29-31 Although the rigid geometry of XDK does not precisely reproduce the disposition of the two carboxylate side chains in hemerythrin (vide infra), the ligand nevertheless binds sufficiently tightly to the oxo-bridged dinuclear Fe(II1) core that assembly of the {Fe2014+unit occurs rapidly on mixing 2 equiv of ferric (24) Beer, R. H.; Tolman, W. B.; Bott, S. G.; Lippard, S. J. Znorg. Chem. 1989. . .-. , -28. -, 4551-4559. --
(25) Beer, R. Ph.D. Thesis, Massachusetts Institute of Technology, 1989. (26) Tolman, W. B.; Bino, A.; Lippard, S. J. J. Am. Chem. SOC.1989, I l l , 8522-8523. (27) Mauerer, B.; Crane, J.; Schuler, J.; Wieghardt, K.; Nuber, B. Angew. Chem., I n f . Ed. Engl. 1993, 32, 289-291. (28) Taft, K. L.; Delfs, C. D.; Papaefthymiou, G. C.; Foner, S.; Lippard, S. J. J. Am. Chem. SOC.1994, 116, 823-832. (29) Armstrong, W. H.; Roth, M. E.; Lippard, S. J. J. Am. Chem. SOC. 1987, 109, 6318. (30) Gorun, S. M.; Lippard, S. J. Znorg. Chem. 1988, 27, 149. (31) McCusker, J. K.; Vincent, J. B.; Schrnitt, E. A,; Mino, M. L.; Shin, K.; Coggin, D. K.: Hagen, P. M.; Huffman, J. C.; Christou, G.; Hendrickson, D. N. J . Am. Chem. SOC.1991, 113, 3012-3021.
J . Am. Chem. SOC.,Vol. 1 1 6. No. 12, 1994
5 199
C2'
I
04
C
34
8
Figure 1. ORTEP plot of the [ F ~ ~ O ( X D K ) ( C H ~ O H ) S ( H cation. ~~)~~+ Thermal ellipsoids are drawn at the 35% probability level, and hydrogen atoms are omitted for clarity. Atoms denoted by numbers with asterisks are generated by a crystallographic mirror plane. Atom 0 2 corresponds to the oxygen of both a water and a methanol ligand disordered across the crystallographic mirror plane that passes through Fel-05-Fe2.
nitrate with H2XDK in methanol solution. The spectroscopic features of the reaction mixture clearly support this conclusion, and are similar to those observed during preparation of the molecular ferric whee1.28 In the present case, however, the dinuclear structure of the complex persists in the solid state, so that crystallization affords multigram quantities of [FezO(XDK)(CH30H)5(H20)](NO&, l(N03)2,in high yield. The isolation of 1 demonstrates that XDK plays a dominant role in directing the chemigry of Fe(II1) in solution. Since no capping ligands were present in the reaction mixture, we conclude that the dicarboxylate ligand is responsible for averting the formation of higher nuclearity products. Compound 1 not only represents the first structurally characterized example of a carboxylatebridged dinuclear iron complex having exclusively monodentate ligands in the terminal sites, but the complex is of particular interest because all of the terminal ligands are exchangeable solvent molecules. This feature makes 1valuable both as a starting material for preparing a series of closely related terminal substitution products and also because it affords an opportunity for the detailed study of terminal ligand exchange processes in solution. Both of these objectives have been met in the present work, as detailed below. Structureof Solid l(NO&4HzO. The X-ray structure of this complex (Figure 1, Table 2) reveals the expected bent {Fe2014+ core bridged by both of the carboxylate groups of XDK, with solvent ligands occupying the six terminal positions. The asymmetric unit in the crystal consists of one half of the molecule, with a mirror plane passing through Fel-05-Fe2. Initial attempts to refine all of the coordinated solvent molecules as methanol led to an unacceptably high thermal parameter for the carbon atom bonded to 0 2 . The electron density was more satisfactorily accounted for with the carbon atom refined at halfoccupancy, which corresponds to one CH3OH ligand (C2*-029 and one H 2 0 ligand ( 0 2 ) disordered across the mirror plane. The formula of cation 1 is therefore [ F ~ z O ( X D K ) ( C H ~ O H ) ~ (H20)]2+. The dinuclear units are linked within the crystal lattice by strong hydrogen bonds between the terminal solvent ligands and the nitrate anions, forming infinite chains. This packing arrangement results in wide channels running through the crystal structure. A significant amount of residual electron density was found in this channel, indicating that the structure is highly solvated. The most satisfactory model that we found to account
Watton et al.
5200 J. Am. Chem. SOC.,Vol. 116, No. 12, 1994 Table 2. Selected Bond Distances 1(N08)y4Hzo0
Fe-Fe Fel-01 Fel-03 Fel-05 Fe 1-0 101 Fe2-02 Fe2-04 Fe2-05 Fel-05-Fe2 01-Fel-01* 0 1-Fel-03 0 1-Fel-0101 01-Fel-05 03-Fel-05 03-Fel-0101 0101-Fel-0101*
(A) and Angles (deg) for
Bond Distances 3.125(3) Fe2-0102 2.083(7) 01-c1 2.070(8) 02-c2 1.789(8) 03-C3 2.034(6) 0444 2.082(7) 0 101-c 101 0102-c 101 2.07(1) 1.784(9) Bond Angles 122.0(5) 02-Fe2-02* 86.1(5) 02-Fe2-04 82.9(3) 02-Fe2-05 86.2(3) 02-Fe2-0 102 95.4(3) 04-Fe2-05 177.7(4) 04-Fe2-0102 85.6(3) 0 102-Fe2-0102* 99.2(4)
2.042(7) 1.42(1) 1.42(2) 1.40(2) 1.32(3) 1.25(1) 1.26(1)
86.4(4) 83.0(3) 94.0(3) 87.9(3) 175.9 (4) 86.0(3) 95.6(4)
a See Figure 1for the atom-labeling scheme. Numbers in parentheses indicate the estimated standard deviations in the last significant digits.
for this electron density corresponds to four water molecules per formula unit. These water molecules form hydrogen-bonded chains throughout the crystal and are associated with the cations by hydrogen bonds to the imide carbonyl groups of the XDK ligand (0103,0104) and to the coordinated H2O molecule (02). The structure of 1illustrates an unusual feature in the binding mode of XDK that has been observed in all subsequently characterized Fe(II1) complexes with this ligand. Whereas complexes with simple carboxylates such as acetate typically exhibit metal binding in the plane of the carboxylate group, the rigid conformation of XDK cannot accommodate such a geometry. Instead, the metal ions are situated 1.20 8, above the plane of the carboxylate groups, as illustrated in Figure 2. The magnitude of this deviation is, to our knowledge, unprecedented for a metalcarboxylate complex. Maximal overlap with the lone pairs of the sp2-hybridized carboxylate oxygen atoms should result in the metal atoms being bound in the plane of the carboxylate oxygen and carbon atoms. A survey of dicarboxylate-bridged structures in the Cambridge Structural Database (CSD) indicates the inplane binding geometry to be highly favored for these complexes, in accord with the results found recently for the coordination of transition metal ions to nonbridging carboxylate groups.32 As illustrated in Figure 2, the geometric distortion can be described by the fold angle (4) across the two oxygen atoms of the carboxylate group. The distance (d)of the Fe atoms from the least-squares plane of the carboxylate atoms is also a useful parameter in describing this structural feature. Values of $J and d for XDK derivatives are given in Table 5 , from which it is clear that the binding geometry is maintained for all of the known diiron(II1)-XDK structures. Such an out-of-plane binding mode might be expected to reduce overlap between the carboxylate lone pairs and the metal orbitals, and hence diminish the bond strength. This effect was observed in a complex containing the chelating dicarboxylate ligand MPDPZ-, where lengthening of the Fe-Oco,- bond occurred in association with a modest deviation ( d = 0.49 A) of the Fe atoms from the carboxylate planes* The carboxylate-Fe(lI1) bond distances in 1 do not suggest reduced orbital overlap, however, the average Fe-Ocoodistance being comparable to the average value found for dicarboxylate bridged complexes in the CSD (vide infra). In addition, the spectroscopic parameters obtained for 1 and its derivatives are all within the ranges determined for other (poxo)(h-carboxylato)diiron(III) complexes, indicating that the unusual carboxylate binding mode has little influence over the bonding or electronic structure of the {Fe2OI4+core. (32) Carrell, C. J.; Carrell, H.L.; Erlebacher, J.; Glusker, J. P.J . Am. Chem. SOC.1988, 110, 8651-8656.
Examination of the structure of 1suggests that the carboxylate binding geometry is largely responsible for the predisposition of XDK to form di- rather than tetranuclear (eq 1)~6~21~29~30c~mplexes with Fe(II1). Because the carboxylate groups are essentially coplanar, rather than approximately perpendicular to one another as normally observed, the Fe-0-Fe core resides in a shallow cleft that is flanked by the methyl groups (C104, Figure 1) of the Kemp’s triacid moieties. The methyl groups thus hinder close approach of other molecules to the FeO-Fe unit. Computer modeling reveals that the binding site is sufficiently occluded that formation of oligomeric complexes by condensation of (Fe2O(XDK)JZ+units would result in highly unfavorable steric interactions between ligands. This property of XDK is analogous to the sequestering of a carboxylate-bridged dimetallic unit within the active site pocket of a protein. In complexes of unconstrained carboxylates, “in-plane” binding (Figure 2) results in the side chains being at a maximal distance from the Fe-0-Fe unit; consequently, even bulky ligands such as pivalate (R = But) do not exert a comparable steric influence over the chemistry of the {FezOj4+core. Structureof 1 inSolution. Although theX-ray crystal structure of 1, together with molecular modeling results,strongly suggested that this complex would not oligomerize, we wished to define more rigorously its composition in solution in order facilitate interpretation of results from reactivity and kinetics studies. Comparison of the 57FeMossbauer spectra of 1 in both the solid state and frozen methanol solution indicates, within theresolution of the technique, the presence of a single species in both cases (Figure S1, supplementary material). The spectroscopic parameters obtained from fitting to both spectra are in excellent agreement, differences in the solid state (6 = 0.55, 0.56 mm s-I and AEQ = 1.67,2.06 mm s-1) and frozen solution (6 = 0.56 mm s-1 and AEQ = 1.87 mm s-1) spectra presumably arising from exchange of the coordinated water ligand for a methanol upon dissolution. The broader lines of the solid state spectrum are best fit to a two-site model, having nearly equivalent values of the isomer shift (6), but slightly different quadrupole splittings (AEQ). The magnitude of AEQ is influenced by the asymmetry of the iron coordination environment, and so the larger value is attributed to the Fe(II1) having the HzO ligand in the solid state. In support of the conclusion from the Massbauer data that the complex remains intact in solution, the UV/visible spectrum of 1 also appears to arise from a single species, and the resonance Raman spectrum in methanol solution exhibits only a single v,,,(Fe-0-Fe) stretching vibration at 532 cm-1. We therefore conclude that, in methanol solution, the only chemistry that takes place is exchange of the coordinated water for a sixth methanol ligand, eq 2.
-
CH>OH
lolo) to be determined by this method. The reaction between 1 and bpy is thus taken to be essentially irreversible under these conditions. Attempts to crystallize the nitrate salt of compound 2 from methanol solution afforded only 3, in which one of the methanol ligands is replaced by a nitrate ion. An X-ray structural analysis was carried out on this product, which unequivocally established its composition. One of the nitrate counterions was severely disordered in the crystal lattice, however, and could not be modeled sufficiently well to warrant presentation of the structure here. Further recrystallization of this material from CH2C12/toluene mixtures gave the doubly nitrate substituted complex 4. An attempt to characterize this compound by X-ray crystallography afforded only a low-angle data set (28,, 35O), since the crystal decayed rapidly with solvent loss even a t -100 OC. The quality of the structure was further compromised by disordered CH2C12 solvent in the lattice. We therefore sought an alternative solvent system from which to grow crystals of 4. Suitable material was obtained by crystdlization from acetonitrile, and a structure determination was carried out. The overall molecular formula of the complex is the same as that for crystals obtained from CH2C12, with the exception of the lattice solvents. In contrast to the product obtained from CHzCl2/toluene, however, the complex crystallized from acetonitrile was obtained with the unusual syn stereochemistry with respect to the Fe-0-Fe plane.39 The structure, presented in Figure 4 and Table 3, has a core geometry similar to that of 1, with the iron atoms of the FeO-Fe unit positioned 1.21 A out of the carboxylate planes of XDK. The Fe-L bond distances and angles are quite similar to those found for a related complex, anti-[Fe2O(ClCH2COz)2(bpy)2(N03)21, 7.11 In particular, thecomplexes show similar structural features, including a lengthening of Fe-L bonds involving ligands trans to Fe-Ob, and Fe-ONo,- compared with those involving ligands in cis positions, although the magnitude of this trans influence is smaller in 4 than in 7. Despite the positioning of the (Fe2OI4+ unit above the plane of the carboxylate ligands, the average Fe0 ~ 0 0bond distance in 4 is statistically identical to that in 7, again indicating that the XDK ligand is tightly bound to the (poxo)diiron(III) center. Reaction of 1 with 1-Methylimidazoleand Solid State Structure of [FezO(XDK)(1-MeIm)6]*+,5. In addition to substitution by bidentate ligands, 1also reacts cleanly with monodentate N-donor ligands. Titration with 1-methylimidazole (1-MeIm) results in replacement of the terminal solvent ligands, giving the dinuclear complex [Fe20(XDK)(1-MeIm)6]2+,5, as the final product. The reaction is conveniently monitored by changes in the visible spectrum (Figure 5), which resemble those for the bpy reaction described above. The lower energy 6AI [4T2](4G) band shifts significantly, to 680 nm, and in this case also decreases in intensity. (39) Rardin, R. L.; Tolman, W. B.;Lippard, S. J. New J. Chem. 1991, 15,
-
417-430.
5202 J . Am. Chem. SOC.,Vol. 116, No. 12, 1994
Watton et al. 4 1
I
3.5 3 2.5 W
2
e8
3
P
c
2
1.5 1
0.5
O I
A
500
@
SJW-C~CH~CN‘ Fe-Fe Fel-01 Fel-02 Fel-0101 Fe 1-020 1 Fel-N1 Fel-N2
550
600
I
650
I
I
700
750
I
J
800
Figure 5. Changes in UV/visible spectra during titration of 1 with 1-methylimidazole in CH30H. Concentration of 1 = 0.13 M.
@+---
(A) and Angles (deg) for
Bond Distances 3.1 52(2) Fe2-0 1 1.796(4) Fe2-05 Fe2-0 102 2.090(5 ) 2.004(4) Fe2-0202 2.069(4) Fe2-N3 2.179(5) Fe2-N4 2.122(5)
A
4-----4--
1.795(4) 2.1 12(5) 2.032(4) 2.045(4) 2.178(5) 2.150(5)
Bond Angles Fe 1-0 1-Fe2 122.7(2) 01-Fe2-05 101.0(2) 01-Fe 1-02 98.5(2) 89.5(2) 01-Fe2-0102 01-Fel-0101 92.9(2) 97.6(2) 01-Fe2-0202 01-Fel-020 1 168.6(2) 92.4(2) 01-Fe2-N3 01-Fel-N 1 97.2(2) 171.7(2) 01-Fe2-N4 01-Fel-N2 79.9(2) 100.8(2) 05-Fe2-0 102 02-Fe 1 - 0 101 166.0(2) 87.8(2) 05-Fe2-0202 02-Fel-020 1 86.7(2) 175.0(2) 05-Fe2-N3 02-Fel-N 1 98.1(2) 05-Fe2-N4 86.7(2) 02-Fe 1-N2 0102-Fe2-0202 100.2(2) 88.2(2) 0101-Fel-0201 9 6 32) 0102-Fe2-N3 91.2(2) 0101-Fe 1-N 1 0102-Fe2-N4 86.3(2) 161.1(2) 0 101-Fe 1-N2 79.3(2) 161.2(2) 0202-Fe2-N3 0201-Fel-N 1 89.6(2) 79.8(2) 0202-Fe2-N4 86.9(2) N3-Fe2-N4 020 1-Fe 1-N2 74.7(2) N 1-Fel-N2 76.1(2) a See Figure 4 for the atom-labelingscheme. Numbers in parentheses indicate the estimated standard deviations in the last significant digits. Moreover, significant increases in intensity are observed for the higher energy transitions, which shift slightly in wavelength. In contrast to the apparently straightforward behavior of 1 when titrated with dipyridyl, the spectral changes observed in the titration with 1-MeIm indicated a somewhat more complex situation. Addition of the first 2 equiv of 1-MeIm resulted in loss of the absorption band at 6 16 nm, which was replaced by a broad feature in the spectrum extending from 500 to 800 nm. This behavior was independent of the concentration of 1, and was interpreted as resulting from substitution of 1-methylimidazole into twoof the terminal coordination sites on the dinuclear complex (eq 4). [Fe20(XDK)(CH30H)6]2+
I
Wavelength (nm)
Figure 4. ORTEP plot of syn- [Fe20(XDK)(bpy)2(N03)2]. Thermal ellipsoidsare drawn at the 40%probability level, and hydrogen atoms are omitted for clarity. Table 3. Selected Bond Distances
I
+ 2 1-MeIm-
[Fe,O(XDK)( 1-MeIm),(CH30H)4]Z+ + 2CH30H (4) Substitution of the remaining four methanol ligands by addition of further equivalents of 1-MeIm (eq 5 ) depended on the
350
26equiv
5equiv 4equiv
450 500 550 600 650 700 Raman shift (cm-’) Figure 6. Resonance Raman spectral changes during titration of 1 with 1-methylimidazole in CH3OH. Concentrationof 1 = 0.10 M. Individual spectra represent a sum of 20 scans. 400
concentration of 1 at which the titration was performed. At
-
+ 4 1-MeIm [Fe,0(XDK)(l-MeIm),]2+ + 4CH30H ( 5 )
[Fe,O(XDK)( 1-MeIm),(CH30H)4]2+
lower concentrations (- 10 mM), the substitution reaction, monitored by the increase in A680, required 30-40 equivof 1-MeIm before it was complete. At higher concentrations (-0.1 M), however, the reaction was complete upon addition of 4 equiv of 1-MeIm. This behavior indicates that the equilibrium constants are substantially less for substitution of the four remaining solvent molecules in 1by 1-MeIm than for the first 2 equiv. Substitution at the terminal sites of 1 therefore separates into two classes having different affinities for 1-MeIm. Since the “high affinity” and “low affinity” sites account for 2 and 4 equiv of 1-MeIm, respectively, the difference in reactivity suggests that the two positions trans to theoxo bridge might constitute the”high affinity” binding sites, with the cis sites having the lower affinity. Interestingly, in a departure from the normal behavior of (0x0)(carboxy1ato)-bridged dinuclear iron(II1) centers, the X-ray structure of S(BPh4)z (Figure 7, Table 4) reveals that there is no lengthening of the average Fe-Nifidamlc bond trans to the Fe-Ob, unit, relative to the Fe-N,i, distances. This result implies comparatively stronger bonding at these positions than is commonly observed. It cannot be attributed to the imidazole donor, since significant lengthening of Fe-Ntransbonds is apparent
(p
- Oxo)bis (p- carboxy1ato)diiron( I l l ) Complexes
J . Am. Chem. SOC.,Vol. 116, No. 12, 1994 5203
c12
c20
C16
adFigure 7. ORTEP plot of the [Fe20(XDK)( 1-MeIm)s]2+cation. Thermal ellipsoidsare drawn at the 40% probability level, and hydrogen atoms are omitted for clarity. Table 4. Selected Bond Distances (A) and Angles (deg) for S(BPh)&HCl3" Fe-Fe Fel-01 Fe 1-0101 Fe 1-0201 Fel-Nl Fel-N3 Fel-N5 Fe 1-0 1-Fe2 0I-Fe 1 - 0 101 01-Fel-020 1 01-Fe 1-N 1 01-Fe 1-N3 01-Fe 1-N5 0101-Fel-0201 0101-Fe 1-N 1 0101-Fel-N3 010 1-Fe 1-N5 0201-Fel-N1 0201-Fel-N3 0 2 0 1-Fel-N5 N 1-Fel-N3 N 1-Fel-N5 N3-Fe 1-N5
Bond Distances 3.226( 1) Fe2-0 1 1.814(4) Fe2-020 1 2.078(4) Fe2-0202 2.050(4) Fe2-N7 2.167(5) Fe2-N9 2.155(5) Fe2-N 1 1 2.157(5) Bond Angles 125.7(2) 01-Fe2-0 102 92.9(2) 01-Fe2-0202 94.0(2) 01-Fe2-N7 173.9(2) 01-Fe2-N9 99.7(2) 01-Fe2-Nl1 87.9(2) 0 102-Fe2-N7 99.8(2) 0102-Fe2-N9 84.6(2) 0102-Fe2-N11 84.9(2) 0202-Fe2-N7 173.6(2) 0202-Fe2-N9 80.9(2) 0 102-Fe2-0202 165.3(2) 0202-Fe2-Nll 86.4(2) N7-Fe2-N9 85.7(2) N7-Fe2-Nll 95.2(2) N9-Fe2-Nll 88.6(2)
1.81l(4) 2.055(4) 2.059(4) 2.146(5) 2.128(5) 2.188(5)
91.1(2) 95.3(2) 173.2(2) 90.2(2) 97.3(2) 82.1(2) 89.5(2) 170.7(2) 85.3(2) 169.6(2) 99.3(2) 84.2(2) 90.3(2) 89.6(2) 86.4(2)
See Figure 7 for the atom-labeling scheme. Numbers in parentheses indicate the estimated standard deviations in the last significant digits.
in structures with bidentate imidazole although the effect is considerably less pronounced in these cases than with pyridine or pyrazole donors. Regardless of its origin, the observation of equivalent Fe-Ntrans distances in 5 lends some support to the notion that the high affinity sites might be those situated trans to the Fe-Ob, bond. Further work is required, however, to establish this assignment definitively. In addition to visible spectroscopy, the reaction of 1 with 1-MeIm (eqs 4 and 5) is amenable to study by resonance Raman spectroscopy, as shown in Figure 6 . The spectral changes reflect differences in the energy of the u,,,(Fe-0-Fe) vibration, which in turn depends on the Fe-Ob,-Fe angle.2 Attempts to monitor the dipyridyl reaction (eq 4) by resonance Raman spectroscopy were unsuccessful, for no change in Raman shift was apparent in the spectra of the reactants and products. Subsequent (40) Sessler, J. L.; Hugdahl,J. D.; Lynch, V.;Davis, B. Inorg. Chem. 1991, 30, 334-336. (41) Taft, K. L.Ph.D. Thesis, Massachusetts InstituteofTechnology,1993.
crystallographic analyses of the products revealed only slight differences in the Fe-0-Fe angles of 1 and the bpy-substituted products 3, anti-4 and syn-4, so that little or no change in the Raman shift would be expected for this reaction. With 1-MeIm, however, addition of successive equivalents to a 0.1 M solution of 1resulted in sequential changes in the spectra which terminated after addition of 6 equiv of ligand. As with the visible spectra, the Raman spectra did not display a monotonic transition between the spectra of 1and 5. Addition of the first equivalent of 1-MeIm shifted the maximum to lower energy, broadened the spectrum, and decreased the maximum intensity. Introduction of the second equivalent resulted in a further shift to low energy, as well as an increase in the intensity of the signal. In accord with the visible titration results, these spectral changes are consistent with substitution of the high-affinity sites on 1, with the broadening and loss of signal intensity upon addition of the first equivalent being due to the formation of a statistical mixture of mono-, di-, and unsubstituted derivatives. Addition of the final 4 equiv of methylimidazole further diminished the vibrational frequency and considerably increased the intensity of the transition. The resonance Raman spectral behavior of a large number of dinuclear (p-oxo)diiron(III)-containing proteins and model complexes has been examined in detail, and the effects of the terminal substitution patterns on the energies and intensities of the spectra have been assessed.2 From this analysis it was concluded that, whereas the presence of unsaturated N-donor ligands in the Fe coordination sphere enhances the Raman scattering intensities, the most pronounced effects result from substitution in the position trans to the oxo bridge. Since there was little change in the Raman intensity after 1 reacted with the first 2 equiv of 1-MeIm, this conclusion is inconsistent with our assignment of the high-affinity substitution sites trans to the FeObr bond. Insufficient data are currently available to resolve this issue. Attempts to characterize structurally derivatives of 1with fewer than 6 equiv of 1-MeIm have not yet been successful. Comparisonof (p-Oxo)diiron(III)XDK Structureswith Those Related Ones in the Cambridge Structural Database. Because it was initially anticipated that the unusual binding geometry of the (Fe20(XDK))4+moiety might affect the Fe-0 bond strengths (and hence the bond distances) in the compounds of XDK, we have compared the metric data for the three complexes studied here with published values on related molecules. A recent statistical survey of the Cambridge Structural Database (CSD)3* indicated that, whereas out-of-plane Fe-carboxylate binding was highly unusual, the Fe-0 distances in the XDK complexes were within the normal range. Data for complexes containing carboxylate-bridged dinuclear centers were deliberately excluded from this analysis, however, rendering it invalid for comparison with the present results. We thereforecarriedout a limited survey of the CSD42to extract the relevant data for comparison with the carboxylate-bridged core of XDK. The database was searched using standard procedures described elsewhere.43 The structures of all complexes containing the (poxo)bis(p-carboxylato)diiron(III) motif were extracted by using the QUESTroutine, whichreturned a total of 25 uniquestructures. The results of our analysis of the relevant geometric parameters, using the GSTAT routine to interrogate the statistical distributions among values, are presented in Table 5. Comparison with the average values found for the X-ray structures of 1, syn-4, and 5 (Table 5) reveals that, with the exception of the parameters 4 and d, which describe the binding geometry of the Fe(II1) atoms with respect to the carboxylate plane (Figure 2), the values for the XDK complexes are in good agreement with those in the survey. We therefore conclude that the unusual positioning of the metal ions in the plane of the carboxylate groups of XDK (42) Allen, F. H.; Kennard, 0.; Taylor, R. Acc. Chem. Res. 1983, 16, 146153. (43) Murray-Rust, P.; Motherwell, S.Acra Crystallogr. 1978,834,25182526.
Watton et al.
5204 J . Am. Chem. SOC.,Vol. 116, No. 12, 1994
Table 5. Comparison of X-ray Structural Data Obtained For Complexes Containing the (Fe20(XDK)#+ Core and Values for Structurally Related Complexes in the Cambridge Crystallographic Database XDK complexes Fe-Oab Fe-Ocmb F*Ntmub F*Ndsb Fe-OmMb FeOdlb Fe-Fe Fe-0-Fe
de 6c
CSD values
1(NOo)z
syn-4
S(BPhd2
mean'
std dev
mean
std dev
no. of values
1.787 2.038
1.796 2.038 2.179 2.136
1.813 2.06 2.157 2.157
0.04 0.02 0.02 0.02
2.101 3.152 122.8 1.21 37.2
3.226 125.7 1.20 36.8
1.798 2.05 2.17 2.15 2.07 2.09 3.17 123.5 1.20 37.0
1.794 2.049 2.206 2.15 2.036 2.131 3.126 122.1 0.246 7.21
0.01 0.05 0.05 0.03 0.03 0.01 0.06 3.5 0.16 4
25 100 48 80 10 2 19 25 100 50
2.070 2.083 3.125 122.0 1.21 37.2
0.01 0.05 2.0 0.07 2.1
Average over all crystallographically independent distances in l(NO3)2, syn-4, and S(BPh&. all such bonds in the complex. See Figure 2 for definitions of these parameters.
For 1, syn-4, and 5, values refer to averages over
140 does not substantially weaken the bonding in the 8 complexes 1, syn-4, and 5. e 120Kinetics of Ligand Exchange for the Substitution of Dipyridyl for Solvent in 1. The kinetics and mechanisms of terminal ligand 100exchange reactions at carboxylate-bridged dinuclear centers in non-heme iron proteins are an important aspect of their reactivity. 80Anion binding and exchange reactions determined for metHr with N3-, SCN-, F-, C1-, and O H have revealed these reactions 60to be comparable in rate to substitution reactions at solvated Fe(III), with rate constants on the order of 102-103 M-l s - ~ . ~ 40Although kinetic data are available for bridge-exchange reactions in one model system,17comparison of the protein reactivities with 20 data obtained for the more relevant terminal ligand exchange process has not been possible since no well-behaved synthetic 0 analogues were available. Recently, kinetic studies of catalase activity in a Hr model compound were carried out, from which it was suggested that the rate-determining step may be substitution of HO2- for a terminal ligand in the (r-oxo)bis(carboxylato)diiron(II1) catalyst.2' It is therefore of interest to obtain Figure 8. Plot of pseudo-first-orderrate constants ( k h )vs [bpy] obtained quantitative information about the terminal ligand exchange for reaction of 1with dipyridyl in CH3OH at a constant 1concentration of 0.025 mM. Temperature = 296 f 2 K. Open and closedcircle-sindicate kinetics in model systems amenable to such an investigation. data obtained at 461 (appearance of 2) and 601 nm (disappearance of Accordingly, since ligand exchange reactions of 1 are readily l ) , respectively. monitored by changes in thevisibleabsorption spectra (vide supra), the kinetics of the reaction of dipyridyl were measured under b y bPY pseudo-first-order conditions using stopped-flow methods. Com1 [Fe,O(XDK)(bpy)(CH,0H),l2+ 2 (7) plex 1was the limiting reagent in all experiments, and the [bpy] ki ki A was allowed to vary from approximately 25- to 125-fold excess. Both the decrease in the intensity a t 601 nm, corresponding to for the clean first-order behavior of the reaction kinetics is that loss of 1, and the appearance of 2, as reflected in the increased kl and k2 are widely different. If kl >> k2, one would not expect absorbance at 461 nm, were monitored. As shown in Figure S 2 to see comparable kinetic behavior for loss of 1 and appearance (supplementary material), the kinetic traces obtained at both of 2 unless intermediate A had the same spectrum as 1. We wavelengths give good fits to rate laws for first-order buildup and consider this hypothesis to be unlikely, especially in view of the decay, respectively. A plot of ln(k,b) vs In( [bpy]) at constant linear dependence of the absorption intensities on added bpy [ l ] is linear, with a slope of 1.17 f 0.22, indicating a first-order (Figure 3). Alternatively, the observed behavior may arise from dependence on [bpy]. Similarly, a first order dependence on [ 11 the situation in which k2 >> kl. The two Fe(II1) sites in the was inferred from a plot of ln(V0) vs ln[l] at constant [bpy], molecule are separated by more than 3 A, however, and no steric which had a slope of 1.13 f 0.06. An overall second-order rate contact occurs between coordinated bpy ligands as revealed by law, eq 6, therefore applies to the reaction. The first-order the X-ray structures of 3 and 4. It is therefore difficult to invoke a mechanism involvinga dramatic rate enhancement for the second rate = k [ l ] [bpy] (6) substitution process as a consequence of the first. A likely reason for the observed first-order kinetics is the special dependence of the rate on [bpy] was confirmed by the linear fit case of two consecutive, irreversible reactions for which kl = 2k2 to a plot of koh vs [bpy] (Figure 8), from which a value of (1.51 and the molar extinction coefficient of the intermediate A is such f 0.04) X 103 M-1 s-l was obtained for the second-order rate that EA = 1/2(tl €2) (eq 7).45 If the individual rate constants constant (k). This result is in agreement with data for terminal for the two steps were identical, non-first-order kinetic traces ligand exchange in hemerythrin." and a non-integral dependence of kob on [bpy] would result. It The reaction of 1with dipyridyl is expected to occur in a stepwise is reasonable to expect, however, that the two Fe(II1) sites in 1 manner (eq 7), with each of the two iron centers undergoing an will behave independently, both kinetically and spectroscopically. independent substitution reaction. From the results obtained for In this case, k2 would be less than kl by a statistical factor of 2, spectroscopic titration of 1 with bpy (Figure 3), the individual since A has only one Fe(II1) site remaining for reaction with bpy. steps of the reaction are effectively irreversible. One possibility
4
-
-
+
(44) Meloon, D. R.; Wilkins, R. G. Biochemistry 1976, 15, 1284-1290.
(45) Wilkins, R. G. Kinetics and Mechanism of Reactions of Transition Metal Complexes, 2nd ed.; VCH: Weinheim, Germany, 1991; p 22.
(p
- Oxo)bis(p- carboxylato)diiron(II)
Complexes
Similarly, the absorption spectrum of A is likely to resemble that of an equimolar mixture of 1and 2. These conditions are precisely the ones which mask the two steps of the reaction and lead to the observation of a single first-order process.4s In this case, the observed rate constant k corresponds to kz, and kl = 2 X k = (3.02 f 0.08) X 103 M-1 SI. The present kinetic results support the previous observation, based on comparisons of the reactivity with simple mononuclear Fe(II1) complexes such as [Fe(H20)6l3+, that substitution reactions a t the active site of hemerythrin proceed with rates similar to those exhibited by small molecules in solution.44 The carboxylate-bridged (p-oxo)diiron(III) center neither substantially diminishes nor enhances the rate of terminal ligand substitution. It should be noted, however, that the terminal substitution rates are several orders of magnitude faster than reactions in which the bridging ligands are exchanged.” This information should be valuable in interpreting the reactions of biological non-heme diiron centers and their model complexes. Summary and Conclusions The major findings of this paper may be summarized as follows: 1. The use of a structurally constrained dicarboxylate ligand derived from studies in molecular recognition has afforded a series of oxo-bridged dinuclear Fe(II1) complexes with unusual stability, including unique species with only monodentate ligands in the terminal, capping positions. 2. In contrast to complexes prepared by using simple carboxylate ligands such as acetate, these derivatives are resistant todimerization (eq 1) and exist as dinuclear species both in solution and in the solid state.
J . Am. Chem. SOC.,Vol. 116, No. 12, 1994 5205
3. Since the solvent0 complex remains dinuclear in solution, detailed study of the reactivity a t the terminal sites of carboxylatebridged dinuclear non-heme iron model complexes was possible. Both visible and Raman spectroscopic methods sensitively reveal features of ligand substitution by dipyridyl or 1-methylimidazole ligands. 4. The kinetics of dipyridyl ligand substitution were investigated, affording a quantitative measure of the rate constant. This work revealed that incorporation of ferric ion into the site found in several non-heme diiron carboxylate proteins does not significantly modify the intrinsic rate of terminal ligand exchange. Acknowledgment. This work was supported by grants from the National Institute of General Medical Sciences (to S.J.L. and J.R.). We are grateful to F. M. MacDonnell and R. H. Holm for Massbauer spectra and to A. Bakac, L. E. Pence, D. P. Goldberg, and K. L. Taft for helpful discussions. A.M. thanks the Human Frontier Science Program Organization for a fellowship and research funding. SupplementaryMaterial Available: Tables of complete atomic positional parameters, anisotropic temperature factors, and bond distances and angles for 1 ( N 0 3 ) ~ 4 H 2 0syn-C3CH3CN, , and S(BPh&CHC13 and Figures S1 and S2 displaying Mossbauer data for 1 in the solid state and in methanol solution and kinetic traces for the reaction of 1 with 2,2’-dipyridyl (49 pages). This material is contained in many libraries on microfiche, immediately follows this article in the microfilm version of the journal, and can be ordered from the ACS; see any current masthead page for ordering information.
